On Fri, 2014-08-22 at 09:45 -0700, Dave Hansen wrote:

On 08/22/2014 09:32 AM, Tim Chen wrote:

quoted

quoted

quoted

+#ifdef CONFIG_DEBUG_OBJECTS_FREE
+	"DEBUG_OBJECTS_FREE",
+#endif
+#ifdef CONFIG_DEBUG_KMEMLEAK
+	"DEBUG_KMEMLEAK",
+#endif
+#ifdef CONFIG_DEBUG_PAGEALLOC
+	"DEBUG_PAGEALLOC",

I think coverage profiling also impact performance.
So I sould also put CONFIG_GCOV_KERNEL in the list.

Would CONFIG_GCOV_PROFILE_ALL be the better one to check?  With plain
GCOV_KERNEL, I don't think we will, by default, put the coverage
information in any files and slow them down.

CONFIG_GCOV_PROFILE_ALL is definitely a no no regarding to
performance impact, which is mentioned in the gcov documentation.

I haven't tested this, but if profiling is turned on only for
a piece of code that is performance critical but not for
the whole kernel, in theory performance can still be impacted
with the overhead.  So I think it is safer to check
for CONFIG_GCOV_KERNEL, that has no reason to be turned on
for any workload that's performance critical.

Tim


--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>